July Volume 9, Number 3 1999 ISSN 0268-0130

THE HERPETOLOGICAL JOURNAL

Published by the Indexed in BRITISH HERPETOLOGICAL SOCIETY Current Contents HERPETOLOGICAL JOURNAL, Vol. 9, pp. 101-110 (1999)

E VARIATION IN MADECASSUS MILLOT & GUIR , 1950, A LITTLE KNOWN MALAGASY , WITH RESURRECTION OF E MANTIDA CTYLUS PAULIANI G UIB , 1974 MIGUEL VENCES' AND FRANK GLA W2

1 Zoo/ogisches Forschungsinstitut und Museum Alexander Koenig, Bonn, Germany

2 Zoo/ogische Staatssammlung, Miinchhausenstr. 21, D-81247 Miinchen, Germany

Based on morphological diffe rences, specimens currently attributed to the Malagasy montane frogspec ies Mantidactylus madecassus can be divided into two distinct groups, which correspond to the geographically separated populations from the mountain massifs of Ankaratra and Andringitra. The Ankaratra populations differ from the Andringitra populations by the lack of distinctly bilobed subarticular tubercles on the fingers; more extended webbing between the toes; and a less contrasting dorsal colour pattern. Furthermore, they are distinguished morphometrically. The name Mantidactyluspauliani Guibe, 1974 is available for the Ankaratra specimens and is hereby resurrected. A lectotype of M. madecassus is designated. The two species share a lack of vomerine teeth and both possess a very short snout, and should be co nsidered as closely related allopatric sister taxa. So far as is known, they occur between 1500 and 2500 m altitude (mainly above 2000 m), in brooks and their tributaries in areas of ericoid vegetation or of ro ck formationswith rupicolous plant co mmunities. A short review of Malagasy montane species is provided, confirming that montane in Madagascar harbours an important diversity of species specialized to high-altitudes. Key words: Anura, Ranidae, Mantellinae, Brygoomantis, montane herpetofauna, Madagascar

INTRODUCTION (see Millot & Guibe, 1950; Blommers-Schlosser & Blanc, which was not collected in the survey of Madagascar contains a rich diversity of habitat 1991), Raxworthy Nussbaum ) , and thus not in­ types, mainly due to the variety of climates. The eastern & (1996b cluded in the list of montane from rainforest belt is separated from the western arid re­ Madagascar published by Raxworthy Nussbaum gions by a high plateau on which special montane & ecosystems are found. The three highest massifs are (1996a). M madecassus is a representative of the most Tsaratanana in the north (2876 m), and Ankaratra speciose and heterogeneous anuran genus in Madagas­ (2642 m) and Andringitra (2658 m) in central Mada­ car. According to the most recent descriptions (Vences gascar. The Malagasy montane herpetofauna is known Glaw Vences, Andreone from extensive collections, harboured mainly in the et al., 1997; & 1997; et al., 1998), the endemic genus Mantidactylus currently con­ Museum National d'Histoire Naturelle (MNHN),Paris, tains described species classified into subgenera but basic information on the biology and ecology of 63 12 (Dubois, Glaw Vences, M most species is still lacking. 1992; & 1994). madecassus is included in the subgenus Brygoomantis Dubois, 1992 Recently, Raxworthy & Nussbaum (1996a) re­ (formerly group) which cur­ viewed the montane amphibian and reptile Mantidactylus ulcerosus rently consists of seven valid species communities of the Malagasy massifs of Andringitra, (Blommers-Schlosser Blanc, Glaw Vences, Ankaratra, and Tsaratanana, based on their own sur­ & 1991, & 1994). Brygoomantis are distinguished from repre­ veys. They found that a relatively large number of sentatives of other subgenera of by a species are restricted to the montane heathland, and re­ Mantidactylus derived karyotype (chromosome number = see jected the hypothesis that this habitat is artificial and 2n 24, Blommers-Schlosser & Blanc, 1991; not known in M faunistically depauperate. In another publication madecassus) and a combination of femoralgland struc­ (Raxworthy & Nussbaum, 1996b), the same authors ture (glands including a prominent rounded structure emphasized the similarities in the montane with ·external median depression, rudimentary glands herpetofauna between the Ankaratra and Andringitra present in females), sexual dimorphism in tympanum massifs. size (males having a larger tympanum than females), One frog species so farknown only fromhigh alti­ slightly distensible single subgular vocal sac in males, tudes of these two massifsis Mantidactylusma decassus slightly enlarged finger and toediscs, semiaquatic and partly diurnal habits, tadpoles with generalized mouthparts and distinct spiral-shaped intestine visible Correspondence: M. Vences, Zoologisches Forschungsinsti­ tut und Museum Alexander Koenig, Adenauerallee 160, through ventral skin, and advertisement call structure D-53 113 Bonn, Germany. Email: [email protected] (series of pulsed calls with low intensity). 102 M. VENCES AND F. GLAW

In the course of our ongoing revisions of the frog Morphometric data were processed statistically with genera of Madagascar, we examined the type material the softwarepackage SPSS forWindows, version 6. 1.2. of Mantidacty/us madecassus and of its junior synonym Samples were compared for representative ratios by M pau/iani. In the present paper, we give a detailed re­ non-parametric Mann-Whitney U-tests. Data were description of both taxa and revalidate M pauliani transformed logarithmically (log ) to render relation­ 10 based on several morphological and morphometric dif­ ships between them linear. A Principal Component ferences. We also provide an updated list of montane Analysis (PCA) was carried out using the log -trans­ 10 amphibians of Madagascar, and a comparison between formed data (three factors extracted). the herpetofaunas of Andringitra and Ankaratra. RESULTS MATERIALS AND METHODS VARIATION IN M. MADECASSUS The following morphological measurements were According to Blommers-Schlosser & Blanc (1991), taken with a calliper to the nearest 0.1 millimeter: SVL M madecassus is known from seven localities: Ankara­ (snout-vent length), HW (head width), HL (head tra, Nosiarivo, Ivangomena, Andohariana, Am­ length), ED (horizontal eye diameter), END (eye-nos­ balamarovandana, and Anjavidilava. The first two lo­ tril distance), NSD (nostril-snout tip distance), NND calities are located in the Ankaratra massif and refer to (nostril-nostril distance), TD (tympanum diameter), the type series of the taxon Mantidactylus pauliani HAL (hand length), FORL (forelimb length), HIL Guibe, 1974 (which was synonymized with M made­ (hindlimb length), FOL (foot length), FOTL (foot cassus by Blommers-SchlOsser & Blanc, 1991), and to length including tarsus). All available specimens of the several ZMA specimens previously also referred to M taxon pauliani were measured, whereas in M pauliani (see Blommers-Schlosser, 1979). All other lo­ madecassus specimens measured were selected accord­ calities are located in the Andringitra massif according ing to their state of fixation, in order to get a sample in a to the map in Blommers-Schlosser & Blanc (1991), comparable state to the pauliani specimens. All meas­ which did not include information on voucher speci­ urements were taken by the same person (MY). mens. We found MNHN.vouchers for all mentioned Institutional abbreviations are as listed in Leviton et al. sites except Ambalamarovandana. Several other locali­ (1985). Webbing formula follows Savage & Heyer ties are corroborated by MNHN specimens and by one (1967) as modified by Myers & Duellman (1982) and ZFMK specimen (Fig. I); all these sites are located Savage & Heyer (1997). To facilitate comparisons with close to each other within the Andringitramassif . other species of Mantidactylus, we also give the for­ Detailed examination and direct comparison of all mula used by Blommers-Schlosser (1979) and most available specimens demonstrated that they can be clas­ subsequent authors who have published accounts on sified into two distinct groups, corresponding to the Madagascan anurans. Andringitra and Ankaratra samples, respectively. Dif­ Femoral glands were examined and photographed ferences are (a) subarticular tubercles on fingers (very under a stereo-microscope. Our description refers to the prominent, mostly bilobed in Andringitra specimens; macroscopic appearance of the gland on the ventral fe­ but single and indistinct in those fromAnka ratra; Fig. mur. In a few specimens, we also carefullyremoved the 2); (b) webbing (more extended in Ankaratra, generally skin of the ventral femur and turned it upside down; by reaching the disc of the fifth toe); and ( d) colouration this procedure, the gland structures, which remain com­ (darker and more uniform in Ankaratra). pletely attached to the skin, could be examined in more Additionally, distinct morphometric differences be­ detail and could easily be distinguished from simple tween both samples were found. Measurements of all granular thigh patches as present in many . available Ankaratra specimens and of a representative

FIG. I. Mantidactylus madecassus (ZFMK 5741 6 from FIG. 2. Ventral side of hands of type specimens of M. Cuvette Boby, Andringitra) in life. madecassus (MNHN 1953.246; left) and M. pauliani (MNHN 1972. 1 508, right). Not to scale. Note difference in shape of subarticular tubercles. TABLE 1. Measurements (in mm) of adult specimens of Mantidactylusmadecassus and M. pauliani. See Materials and Methods section for abbreviations of characters. LT = lectotype, PL T = paralectotype, HT = holotype, PT = paratype.

Status Sex SVL HW HL TD ED END NSD NND FORL FOTL HAL HIL FL M. madecassus

MNHN 1953.246 LT F? 27.5 9.6 9.7 1.9 3.1 1.8 2.2 3.0 16.6 22.4 7.9 47.0 15.3 MNHN 1989.3591 PLT M? 25.4 8.0 8.6 1.6 2.2 1.7 1.8 3.0 12.7 21.3 6.2 36.1 12.0

MNHN 1989.3592 PLT F? 31.0 9.6 10.2 2.3 3.0 1.8 2.0 3.0 16.7 21.0 7.6 43.1 14.5 < MNHN 1989.3594 PLT F? 29.2 10.6 10.3 2.9 3.2 2.0 2.2 3.3 16.3 17.2 7.7 43.7 14.7 > MNHN 1972.1182 F 32.0 10.8 11.l 2.3 3.5 1.9 2.6 3.2 17.2 24.8 8.5 51.0 18.3 �> 10.9 11.0 2.2 3.3 2.1 2.7 3.1 17.1 25.2 8.7 50.5 17.7 ::l MNHN 1972.1185 F 32.4 0 MNHN 1972.1190 M? 27.0 9.2 9.9 1.7 3.1 2.0 2.2 2.8 15.3 21.1 7.0 42.4 14.2 z MNHN 1972.1192 F 33.7 10.9 11.2 2.6 3.5 2.0 2.5 3.4 18.0 25.1 8.4 49.6 16.7 z MNHN 1972. 1198 M 27.0 9.4 9.9 2.0 3.3 1.7 2.4 3.0 15.5 20. l 7.8 45.4 15.0 � � MNHN 1972.1199 M 29.8 9.9 10.7 2.9 3.4 2.0 2.1 3.3 17.6 23.7 8.8 48.6 16.8 ::j 13.1 MNHN 1972. 1204 F 29.3 9.5 10.0 2.0 3.1 1.8 2.2 3.3 16.2 25.3 7.7 44.7 � 31.4 10.5 10.2 2.1 3.5 2.1 2.3 3.4 19.0 21.4 8.5 51.4 17.2 � MNHN 1972.1206 F ..., � M. pauliani � MNHN 1972.1508 HT M 3 J.0 I J J.0 11.0 2.5 3.1 1.8 2.3 2.6 18.0 25.7 8.9 50.4 17.2 � 17.4 24.4 �o 47.5 17.0 MNHN 1972.1509 PT F? 29.7 11.2 11. 7 2.8 3.1 1.9 2.1 3.1 � MNHN 1972.1510 PT F 33.7 11.5 10.8 2.2 4.0 1.6 2.2 3.1 18.7 24.2 8.9 48.7 17.0 Q MNHN 1972.1511 PT F 3l.1 10.7 10.8 2.0 3.3 1.5 2.1 2.7 17.6 23.8 8.3 47. 1 16.0 � MNHN 1972.1512 PT F? 27.7 10.0 10.5 2.2 3.4 1.5 2.1 3.0 17.5 24.2 8.7 42. 1 17.0 � MNHN 1972.1513 PT F? 25.8 9.6 9.7 2.1 3.4 1.7 1.9 2.7 16.6 23.1 8.4 44.9 15.8 MNHN 1972. 1514 PT M 29.5 11.0 11.3 2.7 3.3 1.6 2.1 3.0 17.7 24.5 9.0 47.4 16.7 MNHN 1972. 1515 PT F 32.9 10.6 1 l.1 2.2 3.4 1.6 2.0 3.1 18.2 23.8 9.6 49. J 16.7 MNHN 1972. 1516 PT M? 31.6 11.5 11.9 3.1 3.5 1.7 2.3 3.0 18.9 25.5 8.9 50. 1 17.3 ZMA 6803 (I 184) F 31.6 JO.I 10.4 2.0 3.4 1.4 2.3 3.0 16.5 22. 1 8.1 45.5 15.6 ZMA 6803 (1185) F? 27.5 9.2 9.3 2.3 3.4 1.4 1.7 3. 1 15.7 21.8 7.8 42.8 15.2 14.7 ZMA6803 (I 186) F? 26.9 8.8 9.2 1.6 3.2 1.4 1.8 2.7 16.0 21.4 8.6 44.8 41.6 14.0 ZMA 6803 ( 1187) F? 24.4 8.6 9. 1 1.6 3.1 1.3 1.7 2.8 15.0 20.2 7.4 1.8 2.7 15.4 20.5 8.0 41.1 14.0 0 ZMA6803 ( 1188) M? 24.8 8.8 9. 1 1.8 3.2 1.5 w 104 M_, VENCES AND F. GLAW

TABLE 2. Principal component loadings (PCL) of the first three principal components from a principal component analysis of data in Table I (log-transformed), given separately for analyses of male and female data. The five most influential PCLs for each principal component are marked by a superscript ranking.

PCA of males PCA of females Variable PCLI PCL2 PCL3 PCLI PCL2 PCL3 SVL 0.107 0. 1705 -0.077 0.094 0.093 0.122 HL 0.1 152 0.037 -0.040 0.1 024-5 0.024 0.268 HW 0.1104-5 0.033 -0.076 0. 1071 -0.016 0.123 END 0.005 0.4061 -0.059 0.060 0.4191 -0.2785 NND 0.015 0.2933 0.5471 0.062 0.3632 0.2924 NSD 0.071 0.3012 -0.3 16 0.083 0.2385 -0.6232 ED 0.077 -0.010 0.46 l2 0.085 -0.2883 -0.106 TD 0.105 0.062 0.2623 0.086 0.113 0.8111 FORL 0.1171 -0.050 0.011 0.1024-5 -0.149 0.033 HAL 0.099 -0.2144 0.097 0.092 -0.2634 0.032 HIL 0.1 143 -0.001 -0.1085 0. 102 -0.029 -0.4583 FOL 0.108 -0.118 -0.065 0. 1062-3 -0.087 -0.100 FOTL 0.1104-5 -0.115 -0.1704 0. 1 062"3 -0.099 -0.114

sample of Andringitra specimens are given in Table 1. ses of males and females.After removal of the firstfac­ We performeda statistical comparison of both samples tor (equivalent to the size or growth effect),the second using Mann-Whitney U-tests on the ratios in Table 3; and third factors (equivalent to the most important males and females were tested separately. Males and shape factors) were plotted in Fig. 3. Both samples females from Ankaratra had relatively longer hands were mainly separated along the second principal com­ (P<0.05 in females;P

Females • M. pauliani D M. madecassus Males • M. pauliani o M. madecassus

2.0 ------� 1,5 • • 0 1,5 1,0

0 0 1,0 ,5 • • • • • 0 0 ,5 o.o 0 •

o.o -.5 • "' "' c c � � -,5 -1,0 0 &. &. 0 E 0 8 (ij� -1,0 0 a; -1,5 • c. 0 c. ·c:; ·c:; c: c: ·c: 0 a.. -1.5 ,.__--�---�--�---�-----.& -2,0 >1-----�--�-��·---.---�.---�---1 -2,0 -1,0 o.o 1,0 2,0 3.0 -2,0 -1,5 -1,0 -.5 o.o ,5 1,0 1,5

Principal component 2 Principal component 2

FIG. 3. Scatterplots of second and third principal components of a PCA of log-transformeddata in Table I, analyzed separately for males and females. VARIATION IN MA NTIDACTYLUS MA DECASSUS 105

TABLE 3. Differential characters between Ma ntidactylus madecass11s and M. pau/iani. Variable ratios are given as mean ± SD (minimum and maximum values in parentheses). Data were compared using Mann-Whitney U-Tests. Significant differences between data pairs are marked with asterisks: * P<0.05, ** P<0.005, *** P<0.001.

Species M. madecassus M. pauliani (Andringitra) (Ankaratra) SVL (males) 27.0-29.8 mm 24.8-3 1.6 mm SVL (females) 25.4-33.7 mm 24.4-33.7 mm HAL/SVL (males) 0.27 ± 0.02 (0.26-0.30) 0.30 ± 0.02 (0.28-0.32) HAL/SVL (females) 0.26 ± 0.02 (0.24-0.29) * 0.29 ± 0.02 (0.26-0.33) * FOTL/HIL (males) 0.48 ± 0.02 (0.45-0.50) * 0.51 ± 0.01 (0.50-0.52) * FOTL/HIL (females) 0.49 ± 0.01 (0.47-0.50) 0.50 ± 0.03 (0.48-0.57) END/HL (males) 0.19 ±0.01 (0. 1 8-0.20) * 0.15 ± 0.01 (0.14-0.17) * END/HL (females) 0.18 ± 0.01 (0. 17-0.21) *** 0.15 ± 0.01 (0. 13-0.18) *** NSD/HL (males) 0.21 ± 0.01 (0.20-0.22) 0.20 ± 0.01 (0.19-0.2 1) NSD/HL (females) 0.23 ± 0.02 (0.20-0.25) ** 0.19 ± 0.01 (0.18-0.22) ** NND/HL (males) 0.31 ± 0.02 (0.28-0.33) * 0.26 ± 0.03 (0.24-0.30) * NND/HL (females) 0.31 ± 0.02 (0.28-0.35) * 0.29 ± 0.02 (0.25-0.33) * HL/SVL (males) 0.36 ± 0.01 (0.34-0.37) 0.37 ± 0.01 (0.36-0.38) HL/SVL (females) 0.34 ± 0.01 (0.32-0.37) 0.35 ± 0.02 (0.32-0.39) HW/SVL (males) 0.34 ± 0.02 (0.34-0.36) 0.36 ± 0.01 (0.36-0.37) HW/SVL (females) 0.33 ± 0.01 (0.3 1-0.35) 0.35 ± 0.02 (0.32-0.38) TD/ED (males) 0.74 ± 0.17 (0.55-0.91) 0.77 ± 0.14 (0.56-0.89) TD/ED (females) 0.67 ± 0.07 (0.60-0.77) 0.63 ± 0.11 (0.50-0.90) Colour in preservative mostly beige with dark markings. mostly uniformlydark with indistinct darker markings Subarticular finger distinct, bilobed. indistinct, not bilobed. tubercles Foot webbing ends just below disc on toe 5. reaches disc on toe 5.

Within the genus Mantidactylus (and also other A member of the subgenus Brygoomantis as indicated Malagasy anurans), there is no other example of such by (a) sexual dimorphism in tympanum size (males important morphological differences (especially in the having a relatively slightly larger tympanum than fe­ shape of the subarticular tubercles) between males); (b) only slightly enlarged finger and toe discs; populations included in a single species. Generally, (c) well developed webbing between toes; and (d) pres­ even slight morphological differences between ence of femoral glands including a rounded structure populations are paralleled by bioacoustic (see data in with external median depression, and of rudimentary Glaw & Vences, 1994) or, as in the genus Mantella, by femoral glands in females (verified in ZFMK 57416). genetic differentiation (Vences et al., 1998). Some­ M. madecassus is distinguished fromall other spe­ times, bioacoustically well differentiated cies of Mantidactylus by the presence of distinct, Mantidactylus species are even virtually indistinguish­ bilobed subarticular tubercles on the fingers (never able by morphology (Glaw & Vences, 1994). bilobed in other Mantidactylus); further from all other We consider the differences found between the Mantidactylus except M. pauliani by the unique head Andringitra and Ankaratra populations attributed to shape, with a broadly rounded, very short snout; from Mantidactylus madecassus as sufficient support for a other species of the subgenus Brygoomantis except M. distinction at the specific level. Because the name pauliani by the lack of vomerine teeth; and, addition­ Mantidactylus pauliani is available for the Ankaratra ally, from M. betsileanus, M. biporus and M. alutus by populations, we resurrect this name from the synonymy the great extent of the webbing between the toes. For of M. madecassus. furtherdistinction from Mantidactylus pauliani, see be­ low. REDESCRIPTION OF M. MA DECASSUS AND M. Name bearing type. Lectotype (hereby designated) PA UL/AN! MNHN 1953.246 (Fig. 4). Collected by J. Millot at Cirque Boby, Andringitra massif. Specimen in medio­ MA NTIDACTYLUS MADECASSUS MILLOT & GUIBE, 1950 cre state of preservation. Colour patterns well Diagnosis. A species of the genus Mantidactylus as preserved, dorsally beige with irregular large dark indicated by the lack of nuptial pads in males and the brown markings, ventrally uniformly light. Fingertips presence of femoral glands (verifiedin ZFMK 5741 6). slightly enlarged. Fingers with very distinct bilobed 106 M. VENCES AND F. GLA W

during their 1970/71 expedition to the Andringitra mas­ sif: MNHN 1972. 1181 (Cuvette Boby, 28.11.1970), E 1182-3 (Marositry), 1184-5 (Andohabatomanara, 0 Varavarana, ruisseau, 18.11.1970), 1186 (Plateau Andohariana, riv. Riambavy, ea. 2030 m altitude, 6.12.1970), 1189 (Cuvette Boby, 25 .11.1970), 1190 (Ibory face ouest, 17.12.1970), 1191 (Cuvette Boby), 1192 (Ibory face ouest, 25.11.1970), 1193-5 (Cuvette Boby), 1196-7 (Ibory face nord, ruisseau, 16.12.1970), 1198-9 (Antsifotra, 4.12.1970), 1200- 1 (plateau Andohariana, riv. Riambavy, 2.12.1970), 1202-3 (Anjavidilava, 9. 1.1971 ), 1204-5 (Ibory face sud, 16.12.1970), 1206 (Cirque Boby, 26.11.1970). MNHN 1972. 1 187 (Ibory, 16.12.1970) and 1188 (Ivangomena) were not examined; in the MNHN catalogue, they are FIG. 4. Lectotype of M. madecassus (MNHN 1953.246, left) provided with the comment "echange Duellman 1977" and holotype of M. pauliani (MNHN 1972.1508) in dorsal and almost certainly they correspond to the specimens view. KU 173060-61 (L. Trueb, pers. comm.). subarticular tubercles. No femoralgl ands visible (pos­ All examined specimens correspond well with the sibly due to poor fixation). Webbing formula of foot: I type series. Femoral glands (a small, rudimentary struc­ J+-2· II 1-2 III 1+-2+ IV 2+ - 1 V. Webbing formula ture with median depression and additional irregular according to the notation of Blommers-SchlOsser structures towards the anal region) are recognizable in (1979): 1(0.25), 2i(0.75), 2e(O), 3i(l), 3e(0.25), 4i(l .5), ZFMK 57416 (probably a female),but are not evident 4e(l .25), 5(0). Toe 5 of same length as toe 3. Foot in MNHN specimens, probably due to poor fixationand longer than tibia. Tibiotarsal articulation reaches the preservation. centre of the eye. Distribution. Locality coordinates and altitudes are Other typ es. Eight paralectotypes (MNHN given according to Pauli an et al. ( 1971) and Goodman 1989.3590-3597; originally all subsumed under (I 996a, b ). The species is only known from the MNHN 1953 .246) in a mediocre state of preservation, Andringitra massif and was foundat 9-10 localities be­ partly in worse state than the Jectotype, all with the tween ea. 1500 and 2500 m altitude (most localities same collection dates and locality as the lectotype. The higher than 2000 m): (I) Cirque Boby, 2520 m (see original catalogue entry reads I 0 specimens under Blommers-SchlOsser & Blanc 1991: 254), probably MNHN 1953.246, but only eight specimens (beside the 22°1 l 'S/46°53'E; (2) Cuvette Boby, ea. 2470 m, lectotype, which retained the original number) were 22°1 I 'S/46°53'E; (3) Andohariana plateau, ea. 2030 given new numbers in 1989. The missing specimen is m, 22°09'S/46°54'E; (4) Anjavidilava, 1800-2100 m, probably BMNH 1950.1 .2.80, which is Jabeled as co­ 22°09'S/46°57'E; (5) Marositry, ea. 2000 m, 22°10'S/ type and was received in 1950 from J. Guibe. This 46°56'E; (6) Varavarana 1500-1850 m, 22°08'S/ specimen, in a better state of preservation than the 46°57'E; (7) Antsifotra, ea. 2000 m, 22°10'S/46°56'E MNHN specimens, must therefore also be regarded as (altitudes and coordinates referring to Antsifotra paralectotype. River); (8) Ivangomena, 2100-2500 m, 22°09' SI All paralectotypes agree morphologically with the 46°53'E; (9) Ibory (north, west and south slope), prob­ Jectotype. The followingdata referto variation in three ably referring to Pie Bory, up to 2630 m, 22°12'S/ specimens (MNHN 1989.3591-3592, 1989.3594). 46°55'E; possibly also (10) Ambalamarovandana, alti­ Webbing formula I (I - J+) -(1+ -2) II (1 - J+)-(2 - tude ea. 1530 m, 22°08'S/46°57'E (locality in 2+) III I+ - (2 - 3) IV (2 - 3·)-(1 - I+) V. Webbing for­ Blommers-Schlosser & Blanc 1991, possibly referring mula according to the notation of Blommers-SchlOsser to a personal observation of C. P. Blanc). ( 1979): I (0-0.25), 2i(0.5-l ), 2e(0-0.5), 3i(l-l.5), Natural history. MNHN 1989.3593 (SVL 31.3 mm) 3e(0.25), 4i(l-2), 4e(l-l.75), 5(0-0.5). Toe 5 of same contained 10 very large (diameter 3. 8 mm) uniformly length as toe 3. Foot longer than tibia. Tibiotarsal ar­ yellowish oocytes. MNHN 1972. 1 206 (SVL 31.4 mm) ticulation reaches posterior eye margin in MNHN contained 28 yellowish to light brown oocytes of2-2.3 1989.3591 and 3594, tympanum in MNHN 1972.3592. mm diameter. We foundtwo specimens in a stagnant The prominent bilobed subarticular tubercles on the tributary of a brook, surrounded by ericoid vegetation fingers are visible in all specimens. Femoral glands are (Glaw & Vences, 1994). Specimens were sitting in or not clearly recognizable in any paralectotype, probably near water, and readily dived when disturbed. due to their poor state of preservation. MA NTJDA CTYLUS PA UL/A N! GUIBE, 1974 Additional material examined. ZFMK 57416 (Cu­ vette Boby, Andringitra, collected by F. Glaw and M. Diagnosis. A species of the genus Mantidactylus as Vences on 18 January 1994) and 26 specimens fromthe indicated by the Jack of nuptial pads in males and the MNHN, all collected by C. P. Blanc and co-workers presence of femoral glands. A member of the subgenus VARIATION IN MA NTIDACTYL US MA DECASSUS 107

FIG. 5. Femoral gland of Ma ntidactylus pauliani (MNHN FIG. 6. Femoral gland of Ma ntidactylus pauliani, MNl-IN 1972. 1514, male) in internal view (on the underside of 1972. 151 1, female, in internal view (on the underside of ventral fe mur skin after dissection). Scale = 1 mm. ventral fe mur skin after dissection). Scale = I mm. Brygoomantis as indicated by (a) sexual dimorphism in Other types. Eight paratypes. Four males (MNHN tympanum size (males having a relatively slightly 1972. 1509, 1972.1512, 1972. 1514, 1972. 1516), three larger tympanum than females); (b) only slightly en­ fe males (MNHN 1972. 15 I 0, 1972. 151 1, 1972. 1515), larged finger and toe discs; (c) well developed toe and one possibly subadult male (MNHN 1972.1513). webbing; and (d) presence of femoral glands including MNHN 1972.1509-1513 with the same collecting dates a prominent rounded structure with external median de­ and locality as the holotype; MNHN 1972. 1514-1516 pression, and of rudimentary femoral glandsin fe males. collected by C. P. Blanc at Betay forest, Ankaratra mas­ M. pauliani is distinguished from all other sif. All in excellent state of preservation. Dorsal colour Mantidacty/us except M. madecassus by the unique rather uniform,but lighter with large dark brown mark­ head shape, with a broadly rounded, very short snout; ings in a few specimens (e. g. MNHN 1972.1509). from other species of the subgenus Brygoomantis ex­ The following data refer to variation in three cept M. madecassus by the lack ofvomerine teeth; and, paratype specimens (MNHN1972. 1510, 1972. 1514- additionally, from M. betsileanus, M. biporus and M. 1515). Webbing formula I 1-1+11 1-2 Jll l+-2 IV (l+ a/utus by the great extent of the foot webbing. M. - 2) - o+ V. Webbing formulaaccording to the notation pauliani is distinguished from M. madecassus by the of Blommers-Schlosser (1979): 1 (0), 2i(0.5), 2e(O), single and indistinct subarticular tubercles on the fin­ 3i(l), 3e(0.25-0.5), 4i(l), 4e(0.25-1), 5(0). Toe 5 of gers, by the great extent of the webbing between the same length as toe 3. Foot longer than tibia. Tibiotarsal toes (webbing extending onto disc of fifth toe), and by a articulation reaches the posterior eye margin in MNHN less distinct dorsal pattern. 1972.151'4, the tympanum in the other two specimens. Name bearing type. Holotype MNHN 1972. I 508 Femoral glands are visible in all specimens, also in (Fig. 4). Collected by C. P. Blanc at Nosiarivo, females (although clearly smaller). The gland section Ankaratra massif. Adult male in excellent state of pres­ with externalmedian depression is small in the males as ervation. Dorsal colour uniformly dark brown, with a compared with other Brygoomantis. The smaller, few light elements on femur resulting in slight proximal groups of granules are also present in females, crossband pattern. Ventrally uniformly cream with a but rosette-like structures are less distinct in most fe­ slight fading from ventral to dorsal colour along the males. It is not easy to distinguish externally between flanks. Finger tips barely enlarged. Greyish outer meta­ male and fe male glands. Females are therefore only tarsal tubercle present, rather large and distinct, slim tentatively sexed except MNHN 1972.1511 which was (2.2 x 0.9 mm). dissected (see below). Webbing formula I 1-l+II 1 -2 III 1+-2 IV 1+-0+ Afterdissection of one male (MNHN 1972.1514 ), it V. Webbing formula according to the notation of could be recognized that the gland section externally Blommers-Schlosser (1979): 1(0), 2i(0.5), 2e(O), 3i(l), showing the median depression consists of two median 3e(0.25), 4i(I), 4e(0.25), 5(0). Toe 5 of same length as granules surrounded circularly by seven slightly toe 3. Foot longer than tibia. Tibiotarsal articulation smaller granules, and the proximal smaller irregular reaches the tympanum. structures are groups of several much smaller granules Femoral glands present, consisting of one larger, (Fig. 5). In a female (MNHN 1972.1511), the rudimen­ rounded externally prominent structure with a median tary median depression section was a central granule depression, and irregular, less prominent groups of surrounded by seven granules of similar size, and the granules proximally. A rosette-like structure of several proximal irregular groups of granules were smaller and of these small granule groups is recognizable. more widely spaced (Fig. 6). 108 M. VENCES AND F. GLA W

Additional material examined. ZMA 6803, five similarity of the structure of the femoral glands to cer­ specimens individually labeled with the field numbers tain species of Brygoomantis, such as Mantidactylus 1184-1188, collected by R. Blommers-Sch!Osser on curtus, is striking (pers. abs.), and the subgeneric attri­ 21.3 .1973 at Nosiarivo, Ankaratra massif, 2200 m alti­ bution of M madecassus and M pau/iani is therefore tude. rather certain. Distribution. Known from two collecting sites, lo­ cated close to each other in the Ankaratra massif at RELATIONSHIPS OF THE MONTANE AMPHIBIAN FAUNA altitudes above 2000 m: (1) Nosiarivo, 2200 m (see OF ANKARA TRA AND ANDRINGITRA MASSIFS Blommers-Schlosser 1979); (2) Betay forest (accord­ A recent analysis of montane herpetofaunas in ing to Viette 1991: Ambohimirandrana, 2100/2200 m). Madagascar (Raxworthy & Nussbaum, 1996a) com­ Natural history. One female (MNHN 1972. 1511) pared the high mountain communities endemic to contained 16 mature uniformly yellowish oocytes of altitudes higher than 1500 m of the three highest mas­ 3.5 mm diameter. According to Blommers-Schlosser sifs in Madagascar: Andringitra, Ankaratra and (1979), specimens were foundunder boulders in rapids. Tsaratanana. As these authors found no additional spe­ DISCUSSION cies of amphibians and reptiles during the final 3-5 survey days at each site, they considered the species RELATIONSHIPS OF MA NTIDA CTYLUS MADECASSUS AND lists compiled during their surveys as "nearly com­ M. PA ULIANI plete". Their reptile data actually appear to be relatively Mantidactylus madecassus and M pau/iani can be complete, in comparison with available literature. Ac­ considered as strictly montane amphibian species, oc­ cording to Raxworthy & Nussbaum's (1996a) list, curring between 1500 and 2500 m altitude, mainly however, they only recorded the amphibians above 2000 m. , so faras is known, are brooks Anodonthyla montana, Plethodontohyla tuberata, and their tributaries in areas of ericoid vegetation or of microtympanum, Mantidactylus aerumnalis, rock formations with rupicolous plant communities M alutus, and M domerguei as high mountain species (compare locality lists with site descriptions in Paulian of the Ankaratra and Andringitra massifs. At least half et al., 1971; Blommers-Sch!Osser, 1979; Glaw & of the mentioned species also occur lower than 1500 m Vences, 1994; Goodman l 996a). altitude and therefore cannot be viewed as high moun­ The distinguishing featuresbetween M madecassus tain endemics in the sense defined by Raxworthy & and M pauliani have already been shown by Guibe Nussbaum (1996a). M domerguei is known between (1978) in his drawings of the hands and feet of both 900-1800 m altitude (Glaw & Vences, 1994), and M species (figs. 14-15 versus 72-73; M madecassus hav­ alutus occurs at least down to 1200 m altitude. M ing bilobed to paired subarticular tubercles on fingers, aerumnalis sensu Andreone & Gavetti (1994) is only and its webbing not reaching the disc of the fifth toe). known from thetype locality Andrangoloaka (1389 m Guibe (1978) also mentioned the special subarticular according to Viette, 1991) and An'Ala (ea. 850 m, pers. ' tubercles of M madecassus as "elargis abs. ) and therefore is clearly not a montane species'. transversalement, ceux du doigt 3 parfois doubles". The records of M aerumnalis of Raxworthy & However, neither in the original description of Nussbaum (I996a) may actually refer to M Mantidactylus pauliani (Guibe, 1974) nor in his mono­ brevipalmatus (which in the past was considered as a graph (Guibe, 1978) did he mention the obvious synonym of M aerumnalis; see Blommers-Schlosser & external resemblance of pau/iani and madecassus. Blanc, 1991), but even this species is possibly not a Their similarity was seemingly first noted by high mountain endemic. According to Blommers­ Blommers-Sch!Osser & Blanc (1991). These authors, Schlosser & Blanc (1991) it is known from Mandraka, however, did not recognize the differencesbetween the which lies at about 1200 m altitude. Equally, Boophis tax a (characterizing the subarticular tubercles as "peu microtympanum is possibly not a high mountain en­ saillants", probably based on examination of pauliani demic, but the of populations fromthe lower specimens only) and synonymized them without any altitudes of Andringitra needs further study (see Glaw discussion. & Vences, 1994). Although Plethodontohyla tuberata The head shape shared by Mantidactylus has been recorded from Angavokely (Blommers­ madecassus and M. pau/iani is unique among Mala­ Schlosser & Blanc, 1991) which is at 1400 m (Viette, gasy frogsand may be regarded as a derived state. The 1991), this species and Anodonthyla montana can be rounded, short snout strongly resembles that of Euro­ considered as high mountain endemics, at least in a less pean newts, genus Triturus, in their aquatic phase, and strict sense, according to current knowledge. may be a morphological adaptation to largely aquatic Species which are more likely to be true high moun­ habits. tain endemics were apparently not foundin the surveys Attribution of the two species to the subgenus of Raxworthy & Nussbaum (I99 6a, b ), and thus not in­ Brygoomantis is mainly based on femoral gland mor­ cluded in their accounts. Boophis williamsi is only phology, because the other main synapomorphies of the known from 2200m altitude (Guibe 1974, Blommers­ subgenus (reduced chromosome number and spiral­ Sch!Osser, 1979), whereas B. laurenti occurs shaped intestine of tadpole) are unknown. However, the 1500-2650 m above sea level (Guibe, 1974). The few VARIATION IN MA NTJDA CTYLUS MA DECASSUS 109 exactly known localities for Scaphiophryne madagas­ both massifs (data from Glaw & Vences, 1994; cariensis are between 1530 m (Ambalamarovandana) Raxworthy & Nussbaum, 1996a). Other species such as and 2030 m (Andohariana), and Scaphiophryne pustu­ Plethodontohyla tuberata and Anodonthyla montana losa occurs at similar altitudes. Finally, Mantidactylus are known only from one of the two massifs and have madecassus and M. pauliani appear to be confined to no closely related sister species in the other massif. areas above 1500 m and can therefore be regarded as These species are therefore not helpful in indicating high mountain endemics. any vicariant speciation events between Ankaratra and An updated list of the montane amphibians (defined Andringitra. The same is true for Mantidactylus by a general restriction to altitudes of ea. 1500 m or elegans which occurs in Andringitra up to 2500 m, but higher) occurring on the three Malagasy massifs of An­ is not known from Ankaratra. dringitra, Ankaratra, and Tsaratanana (and partly at The climatic history of the Quaternary in Madagas­ other localities of high altitude; see G law & Vences car included dynamic shifts between drier and more 1994) is therefore as follows (data from Raxworthy & humid periods (Battistini, 1996). According to the data Nussbaum, l 996a; Andreone et al., 1998; and as dis­ summarized in Burney (1996), ericoid vegetation of cussed above): Boophis laurenti (Andringitra), B. Madagascar's highest mountain ranges may have flour­ williamsi (Ankaratra), B. ankaratra (Andringitra, An­ ished at times during the Pleistocene down to elevations karatra), Mantidactylus madecassus (Andringitra), M. of ea. I OOO m. This continuous montane habitat was pauliani (Ankaratra), Anodonthyla montana (Andringi­ probably populated by the ancestor of the M. tra), Plethodontohyla tuberata (Ankaratra), P. madecassus/pauliani clade, as well as by the ancestors guentherpetersi (Tsaratanana), P. sp. A (Tsaratanana), of the other allopatric sister species. Due to a period of P. sp. B (TSaratanana), P. sp. C (Tsaratanana), Platype­ warmer climate, it is possible that such high mountain lis tsaratananaensis (Tsaratanana), P. sp. A habitats and their associated amphibian fauna became (Tsaratanana), Stumpffia sp. A (Tsaratanana), Scaphio­ restricted to the high altitude refuges of Ankaratra and phryne madagascariensis (Andringitra), S. pustulosa Andringitra. In this scenario, the subsequent tempera­ (Ankaratra), and S. sp. A (Tsaratanana). Additional ture decreases were not intense enough to allow the species which may be montane endemics but which montane habitats of both massifs to get into long-term need taxonomic revision are Boophis microtympanum contact at lower altitudes again; hence, montane (Andringitra, Ankaratra), Mantidactylus brevipalmatus endemics of both massifs, such as the Mantidactylus (Andringitra, Ankaratra), and Mantidactylus elegans madecassus and M. pauliani ancestor, remained iso­ (Andringitra, Tsaratanana). This list contains up to 17, lated and several of them evolved into separate species. possibly 20 species, whereas Raxworthy Nussbaum & ACKNOWLEDGEMENTS (1996a) listed 14 montane amphibian species. The gen­ eral conclusion drawn by these authors - that the We are grateful to Annemarie Ohler and Alain montane heathlands of Madagascar harbour a relatively Dubois (MNHN, Paris), Berthus van Tuij l (ZMA, Am­ diverse endemic amphibian fauna- is thereforestrong­ sterdam) and Barry Clarke (BMNH) for making ly confirmed. material accessible. Linda Trueb (KU, Kansas) pro­ Three species pairs of montane amphibians may be vided information on specimens held in her care. The examples of vicariant speciation between the fi eldwork of FG was financially supported by Andringitra and Ankaratra massifs. Mantidactylus "Deutscher Akademischer Austauschdienst" (DAAD). madecassus (Andringitra) and M. pauliani (Ankaratra) The work of MV was supported by a grant from the appear to be sister species due to the probably "Graduiertenforderung" of the Nordrhein-Westfalen synapomorphic head shape and lack of vomerine teeth. state. The research in Madagascar was made possible Scaphiophryne madagascariensis (Andringitra) and S. by cooperation between the University of Antananarivo pustulosa (Ankaratra) share a greenish colouration with and the "Zoologisches Forschungsinstitut und Museum sharply bordered brown markings which are not known A. Koenig", Bonn. We are indebted to the Malagasy in other Scaphiophryne (see illustrations in Glaw & authorities for research permits, and the permits to ex­ Vences, 1994). Boophis laurenti (Andringitra) and B. port voucher specimens. williamsi (Ankaratra) share a large relative foot length REFERENCES and probably large blackish tadpoles (data in Blommers SchlOsser & Blanc, 1991; Glaw & Vences, 1994). An Andreone, F. & Gavetti, E. ( 1994 ). On the identity of additional example is found in reptiles: the geckos Mantidactylus aerumnalis (Peracca, 1893) (Anura: Lygodactylus intermedius (Andringitra, 1700-2600 m, ). Boll. Mu s. reg. Sci. nat. Torino 12, 57- see Pasteur, 1995) and L. mirabilis (Ankaratra, 2300- 71. 2640 m). Andringitra and Ankaratra populations of all Andreone, F., Glaw, F., Vences, M. & Vallan, D. ( 1998). other amphibian and reptile species which can be con­ A new Mantidactylus fromsouth- eastern Madagascar, sidered as mountain endemics (Boophis ankaratra, with a review of Mantidactylus peraccae (Ranidae: Phelsuma barbouri, Calumma hilleniusi, Furcifer Mantellinae). Herp. J. 8, 149- 1 59. campani, Amphiglossus sp., Ma buya boettgeri, Battistini, R., ( 1996). Paleogeographie et variete des Mabuya madagascariensis) appear to be conspecificon milieux naturels a Madagascar et dans Jes lies voisines: 110 M. VENCES AND F. GLAW

quelques donnees de base pour l 'etude Pasteur, G. (1995). Biodiversite et reptiles: diagnoses de biogeographique de la «region malgache». In: Actes du sept nouvelles especes fossiles et actuelles du genre de Colloque International Biogeographie de lezards Lygodactylus (Sauria, Gekkonidae) Dumerilia Ma dagascar. 1-17. Lourenco, W. R. (ed.). Paris: 2, 1-21. Societe de Biogeographie - Museum. Paulian, R., Betsch, J. M., Guillaumet, J. L., Blanc, C. & Blommers-Sch!Osser, R. M. A. (1979). Biosystematics of Griveaud, P. (1971). RCP 225. Etudes des the Malagasy frogs. I. Mantellinae (Ranidae). ecosystemes montagnards dans le region malgache. I. Beaufo rtia 352, 1-77. Le massif de l 'Andringitra. 1970- 1971. Blommers-Schlosser, R. M. A. & Blanc, Ch. P. (1991). Geomorphologie, climatologie et groupements Amphibiens (p remiere partie). Faune de Madagascar vegetaux. Bulletin de la Societe d 'Ecologie 2, 198- 75, 1-384. 226. Burney, D.A. (1996). Climate and fire ecology as factors Raxworthy, C. J. & Nussbaum, R. A. (1996a). Montane in the Quaternary biogeography of Madagascar. In: amphibian and reptile communities in Madagascar. Actes du Colloque International Biogeographie de Conserv. Biol. 10, 750-756 Madagascar. 49-58. Lourenco, W. R. (ed.). Paris: Raxworthy, C. J & Nussbaum R. A. (1996b). Amphibians Societe de Biogeographie - Museum. and Reptiles of the Reserve Naturelle Integrale Dubois, A. (I 992). Notes sur la classification des Ranidae d 'Andringitra, Madagascar: A Study of Elevational (Amphibiens Anoures). Bull. mens. Soc. !inn. Lyon 61, Distribution and local Endemicity. In: A Floral and 305-352. Fauna! Inventory of the Eastern Slopes of the Reserve Glaw, F. & Vences, M. (1994). A fieldguide to the Nciturelle Integrate d 'Andringitra, Madagascar: With amphibians and reptiles of Madagascar, second Reference to Elevational Va riation. (Zoology - New edition. Koln (Cologne): Vences & Glaw Verlag. Series), 158-179. Goodman, S.M. (Ed.). Chicago: Glaw, F. & Vences, M. (1997). Neue Daten Uber die Fieldiana 85. Mantidactylus-Vntergattung Sp inomantis aus Savage, JM. & Heyer, W. R. (1967). Variation and Madagaskar, mit Beschreibung einer neuen Art. distribution in the tree frog genus Phyllomedusa in Salamandra 32, 243-258. Costa Rica, Central America. Beitr. Neotrop. Fauna 5, Goodman, S. ( l 996a). Description of the 1993 biological 1I1-131. inventory of the Reserve Naturelle Integrale Savage, J. M. & Heyer, W. R. (1997). Digital webbing d'Andrin gitra, Madagascar. In: A Floral and Fauna! formulae for anurans: a refinement. Herpetological In ventory of the Eastern Slopes of the Reserve Review 28, 131. Naturelle Integrate d'Andringitra, Madagascar: Wi th Vences, M., Glaw, F. & Andreone, F. (1997). Description Reference to Elevational Variation. (Zoology - New of two new frogs of the genus Mantidactylus from Series), 1-6. Goodman, S.M. (Ed.). Chicago: Fieldiana Madagascar, with notes on Mantidactylus klemmeri 85. (Guibe, 1974) and Mantidactylus webbi (Grandison, Goodman, S. (I 996b). Gazeteer of localities mentioned in I 953) (Amphibia, Ranidae, Mantellinae). Alytes 14� the text. In: A Floral and Fauna! Inventory of the 130-1 46. Eastern Slopes of the Reserve Nature/le Integrate Vences, M., Hille, A. & Glaw, F. (1998). Allozyme d'Andringitra, Madagascar: With Reference to differentiation in the genus Mantella (Amphibia: Elevational Va riation. (Zoology - New Series), 306- Anura: Mantellinae). Folia Zoo!. 47, 261-274. 308. Goodman, S.M. (Ed.). Chicago: Fieldiana 85. Viette, P. (1991). Principales localites oil des insectes ont Guibe, J. (1974). Batraciens nouveaux de Madagascar. ete recueillis a Madagascar. Faune de Madagascar, Bull. Mus. natn. Hist. nat. , ser. 3, 171, 1169-1 192. suppl. 2, 1-88. Guibe, J. (1978). Les Batraciens de Madagascar. Bonn. zoo!. Monogr. 11, 1-140. Leviton, A. E., Gibbs Jr., R. H., Heal, E. & Dawson, C. E. (1985). Standards in herpetology and ichthyology: Part I. Standard symbolic codes for institutional resource collections in herpetology and ichthyology. Copeia 1985, 802-832. Millot, J. & Guibe, J. (1950). Les batraciens du nord de I'Andring itra (Madagascar). Mem. Inst. scient. Madagascar (A) 7, 1-36. Myers, C. W. & Duellman, W. E. (1982). A new species of Hy /a fromCerro Colorado, and other tree frog records and geographical notes from western Panama. Am. Mus. Novitates 2752, 1-32. Accepted: 2.2. 99